TRAINING AI TO DETERMINE THE GENDER OF THE MAKERS OF FINGER FLUTING ON CAVE WALLS:

Finger fluting at Gargas Cave, France. Image from Clottes, 2002.

We are all probably aware of the existance finger fluting in caves, it is found all over the world, but it has always been somewhat peripheral to the subject of cave art itself. It is, however, purposeful markings made by people on the cave walls so it needs to be covered in any consideration of cave art. Various examples have been attributed to Neandertals, as well as Homo sapiens men, women and children. Now, a team in Australia is using artificial intelligence to try to clarify the makers of these marks.

“Flutings have the potential to reveal information about age, sex, height, handedness and idiosyncratic markmaking choices among unique individuals who form part of larger communities of practice. However, previous methods for making any determination about the individual artist from finger flutings have been shown to be unreliable4. Accordingly, we propose a novel digital archaeology approach to begin understanding this enigmatic form of rock art by leveraging machine learning (ML) as a tool for uncovering patterns from two datasets, one tactile and one virtual, collected from a modern population. We aimed to determine whether ML can reveal subtle differences in the sex of the artist based on their finger-fluted images.” (Jalandoni et al. 2025:1) In other words they will attempt to have machine learning programs learn to distinguish information like gender and age by analyzing finger fluting created by volunteers. If successful, this could then be applied to finger fluting in cave walls to learn more about the persons who originally created the marks.

Finger fluting attributed to Neandertals, Noire Valley, France. Image from Jean Claude Marquet.

“Experiments were conducted - both with adult participants in a tactile setup and using VR headsets in a custom-built program – to explore whether image-recognition methods could learn enough from finger fluting images made by modern people to identify the sex of the person who created them.” (Lock and Egan 2025:1) The team had participants actually make finger flutings in clay as well as virtually while being videotaped. “Two controlled experiments with 96 adult participants were conducted with each person creating nine flutings twice: once on a moonmilk clay substitute developed to mimic the look and feel of cave surfaces and once in virtual reality (VR) using Meta Quest 3. Images were taken of all the flutings, which were then curated and two common image-recognition models were trained on them. (Lock and Egan 2025:1-2)

Disappointingly, the tests did not produce reliable results. “The VR images did not yield reliable sex classification; even when accuracy looked acceptable in places, overall discrimination and balance were weak. But the tactile images performed much better. ‘Under one training condition, models reached about 84% accuracy, and one model achieved a relatively strong discrimination score.’ Dr. Tuxworth said. However, the models did learn patterns specific to the dataset; for example, subtle artifacts of the setup, rather than robust features of fluting that would hold elsewhere, which meant there was more work to be done.” (Lock and Egan 2025:1-2) Doctor Gervase Tuxworth is one of the experimental team that conducted this study. His statement suggests that the test results were highly variable.

Paleolithic finger fluting from Rouffignac Cave, France. Online image, public domain.

“Overall, the deep learning models achieved high accuracy during training, with AUC values exceeding 0.85 for certain tactile image conditions. These results suggest that the models effectively learned patterns within the tactile dataset and demonstrated strong discrimination between male and female-generated finger fluting images. However, the relatively lower AUC values for virtual images, coupled with their unstable test accuracy, indicate that they do not provide sufficiently distinct features for reliable sex classification. This discrepancy highlights the greater robustness of tactile images over virtual images in capturing relevant classification features. Despite the promising performance on tactile images, deep learning models exhibited a pronounced disparity between training and test performance. While training accuracy consistently increased, reaching near-perfect levels in the later epochs, test accuracy remained unstable and showed no substantial improvement over time. This pattern indicates overfitting, where the models effectively learn dataset-specific features but fail to generalize to unseen test data.” (Jalandoni et al. 2025:10) I find the previous paragraph somewhat confusing. It states “accuracy consistently increased, reaching near-perfect levels” and “accuracy remained unstable and showed no substantial improvement” in two contiguous sentences. In any case, the team did not get reliable results.

Finger fluting made by children, Rouffignac Cave, France. Online image, public domain.

There are a number of possible sources of inaccuracy in the test results.“The instability in test accuracy further suggests that the models struggle to extract robust and generalizable patterns from the finger fluting images, ultimately limiting their reliability for sex classification. A possible contributing factor to this challenge could be individual variation in hand size and fluting characteristics. For example, some females may have larger hands and exhibit stronger fluting patterns resembling those of males, while some males may have smaller hands and display lighter, less pronounced fluting strength. This variability could confuse the model, making it difficult to accurately differentiate between sexes and ultimately hindering its performance on the test set. These results underscore the critical need to increase the dataset size to alleviate overfitting and improve the model’s generalizability. Moreover, the inherent variability in finger fluting images may impose fundamental limitations on the feasibility of using deep learning for sex classification, suggesting that alternative approaches or additional contextual data may be necessary to enhance classification accuracy. The limited success of the tactile data in sex prediction underscores the importance of material-based approaches in understanding finger flutings. While the VR data failed to provide useful results, it opens up new and exciting possibilities for exploring the dynamic aspects of fluting and artistic intent in the future. While a modest achievement, this study highlights the potential of ML to enhance traditional archaeological methods”. (Jalandoni et al. 2025:10) Not every try is guaranteed success.

Koonalda Cave finger flutings, Australia. Photograph Robert Bednarik, 1979.

So, this test did not manage to display reliable accuracy, too many variables in the creation of finger fluting seemingly overwhelmed the software. Also, the experiment apparently did not include children, and it is thought that much finger fluting, at least in European cave contexts, was created by children. If successful, this project would have been a really wonderful development but, alas, it was not. Better luck next time.

NOTE: Some images in this column were retrieved from the internet with a search for public domain photographs. If any of these images are not intended to be public domain, I apologize, and will happily provide the picture credits if the owner will contact me with them. For further information on these reports you should read the original reports at the sites listed below.

REFERENCES:

Andrea Jaladoni, Robert Haubt, Calum Farrar, Gervase Tuxworth , Zhongyi Zhang , Keryn Walshe and April Nowell, 2025, Using digital archaeology and machine learning to determine sex in finger flutings, Scientific Reports, 15:34842. https://doi.org/10.1038/s41598-025-18098-4. Accessed online 12 October 2025.

Lock, Lisa, and Robert Egan, 2025, VR experiments train AI to identify ancient finger-fluting artists, 16 October 2025, The GIST, by Griffith University, https://phys.org/news/2025-10-vr-ai-ancient-finger-fluting.html.

TRAINING AI TO DETERMINE THE GENDER OF THE MAKERS OF FINGER FLUTING ON CAVE WALLS:

Finger fluting from Gargas Cave, France. Photograph 2002 by Jean Clottes.

A couple of weeks ago I wrote again about determining the gender of the maker of a handprint by ratios of finger lengths. Well, staying with the hand, this column is about a project that attempted to train a maching learning (ML) program to determing the gender of the makers of finger fluting. We are all probably aware of finger fluting in caves, it is found all over the world, but it has always been somewhat peripheral to the subject of cave art itself. It is, however, purposeful markings made by people on the cave walls so it needs to be covered in any consideration of cave art. Various examples have been attributed to Neandertals, as well as Homo sapiens men, women and children. Now, a team in Australia is using artificial intelligence to try to clarify the makers of these marks.

Finger fluting believed to be by children, Rouffignac Cave, France. Internet image, public domain.

“Flutings have the potential to reveal information about age, sex, height, handedness and idiosyncratic markmaking choices among unique individuals who form part of larger communities of practice. However, previous methods for making any determination about the individual artist from finger flutings have been shown to be unreliable4. Accordingly, we propose a novel digital archaeology approach to begin understanding this enigmatic form of rock art by leveraging machine learning (ML) as a tool for uncovering patterns from two datasets, one tactile and one virtual, collected from a modern population. We aimed to determine whether ML can reveal subtle differences in the sex of the artist based on their finger-fluted images.” (Jalandoni et al. 2025:1) In other words they will attempt to have machine learning programs learn to distinguish information like gender and age by analyzing finger fluting created by volunteers. If successful, this could then be applied to finger fluting in cave walls to learn more about the persons who originally created the marks.

Neanderthal finger fluting, Noire Valley, France. Photograph by Jean Claude Marquet.

“Experiments were conducted - both with adult participants in a tactile setup and using VR headsets in a custom-built program – to explore whether image-recognition methods could learn enough from finger fluting images made by modern people to identify the sex of the person who created them.” (Lock and Egan 2025:1) The team had participants actually make finger flutings in clay as well as virtually while being videotaped. “Two controlled experiments with 96 adult participants were conducted with each person creating nine flutings twice: once on a moonmilk clay substitute developed to mimic the look and feel of cave surfaces and once in virtual reality (VR) using Meta Quest 3. Images were taken of all the flutings, which were then curated and two common image-recognition models were trained on them. (Lock and Egan 2025:1-2)

Additional finger fluting from Rouffignac Cave, France. Internet image, public domain.

Disappointingly, the tests did not produce reliable results. “The VR images did not yield reliable sex classification; even when accuracy looked acceptable in places, overall discrimination and balance were weak. But the tactile images performed much better. ‘Under one training condition, models reached about 84% accuracy, and one model achieved a relatively strong discrimination score.’ Dr. Tuxworth said. However, the models did learn patterns specific to the dataset; for example, subtle artifacts of the setup, rather than robust features of fluting that would hold elsewhere, which meant there was more work to be done.” (Lock and Egan 2025:1-2) Doctor Gervase Tuxworth is one of the experimental team that conducted this study. His statement suggests that the test results were highly variable.

“Overall, the deep learning models achieved high accuracy during training, with AUC values exceeding 0.85 for certain tactile image conditions. These results suggest that the models effectively learned patterns within the tactile dataset and demonstrated strong discrimination between male and female-generated finger fluting images. However, the relatively lower AUC values for virtual images, coupled with their unstable test accuracy, indicate that they do not provide sufficiently distinct features for reliable sex classification. This discrepancy highlights the greater robustness of tactile images over virtual images in capturing relevant classification features. Despite the promising performance on tactile images, deep learning models exhibited a pronounced disparity between training and test performance. While training accuracy consistently increased, reaching near-perfect levels in the later epochs, test accuracy remained unstable and showed no substantial improvement over time. This pattern indicates overfitting, where the models effectively learn dataset-specific features but fail to generalize to unseen test data.” (Jalandoni et al. 2025:10) I find the previous paragraph somewhat confusing. It states “accuracy consistently increased, reaching near-perfect levels” and “accuracy remained unstable and showed no substantial improvement” in two contiguous sentences. In any case, the team did not get reliable results.

Finger fluting in Koonalda Cave, Australia, Photograph 1979, by Robert Bednarik. 

There are a number of possible sources of inaccuracy in the test results. “The instability in test accuracy further suggests that the models struggle to extract robust and generalizable patterns from the finger fluting images, ultimately limiting their reliability for sex classification. A possible contributing factor to this challenge could be individual variation in hand size and fluting characteristics. For example, some females may have larger hands and exhibit stronger fluting patterns resembling those of males, while some males may have smaller hands and display lighter, less pronounced fluting strength. This variability could confuse the model, making it difficult to accurately differentiate between sexes and ultimately hindering its performance on the test set. These results underscore the critical need to increase the dataset size to alleviate overfitting and improve the model’s generalizability. Moreover, the inherent variability in finger fluting images may impose fundamental limitations on the feasibility of using deep learning for sex classification, suggesting that alternative approaches or additional contextual data may be necessary to enhance classification accuracy. The limited success of the tactile data in sex prediction underscores the importance of material-based approaches in understanding finger flutings. While the VR data failed to provide useful results, it opens up new and exciting possibilities for exploring the dynamic aspects of fluting and artistic intent in the future. While a modest achievement, this study highlights the potential of ML to enhance traditional archaeological methods”. (Jalandoni et al. 2025:10) Not every try is guaranteed success.

So, this test did not manage to display reliable accuracy, too many variables in the creation of finger fluting seemingly overwhelmed the software. Also, the experiment apparently did not include children, and it is thought that much finger fluting, at least in European cave contexts, was created by children. If successful, this project would have been a really wonderful development but, alas, it was not to be. Better luck next time.

NOTE: Some images in this column were retrieved from the internet with a search for public domain photographs. If any of these images are not intended to be public domain, I apologize, and will happily provide the picture credits if the owner will contact me with them. For further information on these reports you should read the original reports at the sites listed below.

REFERENCES:

Andrea Jaladoni, Robert Haubt, Calum Farrar, Gervase Tuxworth , Zhongyi Zhang , Keryn Walshe and April Nowell, 2025, Using digital archaeology and machine learning to determine sex in finger flutings, Scientific Reports, 15:34842. https://doi.org/10.1038/s41598-025-18098-4. Accessed online 12 October 2025.

Lock, Lisa, and Robert Egan, 2025, VR experiments train AI to identify ancient finger-fluting artists, 16 October 2025, The GIST, by Griffith University, https://phys.org/news/2025-10-vr-ai-ancient-finger-fluting.html.

 

FINGER LENGTH AND GENDER IN PAINTED HANDPRINTS:

Apparent female handprint, Pech Merle Cave, France. Image from Pinterest.

On August 5, 2009, I posted a column on Hand prints in Rock Art in which I discussed the fact that a viewer can sometimes determine the gender of a rock art creator by measuring the relative length of the first and third fingers in a hand print. Statistically more males have a longer third finger while more females have a longer first finger.

Handprint, Pech Merle Cave, France. Image from Wikipedia.

Rebecca Coffey wrote in Scientific American (2012) that “In men the index finger is usually shorter than the ring finger, but in most women it’s the other way around, although in some women the fingers are of equal length. In mice the digit ratio corresponds to the female-male hormonal balance in the womb during the week digits form; androgen apparently produces a longer ring finger. Researchers study these ratios to see if they can serve as markers for certain human attributes. So far in 2012, studies have found that girls with a masculine ratio do not get lost as easily; that a feminine ratio in heterosexual girls is associated with bulimia; and that boys with more masculine ratios have more typically masculine facial features.” (Coffee 2012:19)

Handprints, Maltravieso Cave, Spain. Internet image, public domain.

According to A’ndrea Elyse Messer (2013) of Penn State University “the assumption has been that hand prints, whether stencils – paint blown around the hand – or actual paint-dipped prints, were produced by men because other images on cave walls were often hunting scenes. The smaller handprints were assumed to be adolescent boys. Dean Snow, emeritus professor of anthropology, came across the work of John Manning, a British biologist who about 10 years ago tried to use the relationships of various hand measurements to determine not only sex, but such things as sexual preference or susceptibility to heart disease. Snow wondered if he could apply this method to the handprints left in cave sites in France and Spain. ‘Manning probably went way beyond what the data could infer, but the basic observation that men and women have differing finger ratios was interesting,’ said Snow. ‘I thought here was a neat little one off science problem that can be solved by applications of archaeological science.’” (Messer 2013) As it turns out there are numerous ancient handprints in rock art.

Handprints, El Castillo, Spain, image from donsmaps.com, photograph by  Pedro Saura.

An in-depth study of handprints in El Castillo Cave in Cantabria, Spain has provided conclusions about the genders of the makers of the handprints. “Several attempts have been made to develop a system to determine the gender of prehistoric artists with the handprints found in many caves with rock art of this chronology. One of the most prominent attempts, as mentioned above, was by Dean R. Snow. In 2006, Snow studied the hands in Les Combarelles, Font-de-Gaume and the Abri du Poisson with the result that four out of six hands belonged to women. In 2010, Snow along with other authors such as Wang used a computer image method to determine the gender of hand stencils.” (Ravazo-Rodriguez  et al., 2017:378) I find it difficult to imagine a reliable scientific result. With individual variation being such an unknown factor it would seem that the best we can do is use it to make educated guesses, which is pretty much good enough for art historians.

Enhanced handprints, Maltravieso Cave, Spain. Internet image, public domain.

However, a team in Spain determined to study this proposition and then applied the results to Paleolithic handprints found in the cave of El Castillo. “In the experiment, 77 samples (hand stencils) of western adults from the Iberian Peninsula, 46 women and 31 men, were taken. For each modern individual (22 women and 18 men), both the stencils and the real size of their hands were measured. This data was then compared with the Paleolithic stencils to determine whether there was a range of variation between the negative image and the actual hand. The measurements taken into account were the general hand length, index finger length and ring finger length. Discriminatory statistical analysis was used for the experimental work and the measurements collected in the field. In the data obtained in the experimental study, significant differences were observed in the length of male and female fingers, but not in the ring fingers themselves. Discrimina(ting) analyses show that it is the absolute finger lengths and not the ring fingers that are able to discriminate between men and women.” (Ravazo-Rodriguez  et al., 2017:1) I believe that what Ravazo-Rodriguez et al. are saying here is that a simple comparison of ring finger length is not enough, the ratio of first to third fingers must be compared. This only makes sense as we now know that hand prints in caves were made by men, women, adolescents and children so hand sizes, and this finger lengths, vary wildly.

Handprints, Cave of the Hands, Big Sur, California. Image by Esselen Institute.

The results obtained by the Spanish team were actually pretty good. “By applying this function to 21 stenciled hands in El Castillo Cave, it was found that 11 belong to women and 10 to men, indicating equal gender representation. Three of the 21 hands may be wrongly sexed according to the discriminant function. However, there is a significant difference between the real finger measurements and the measurements of their stencils in the experimental study, as the negative images overestimate the real values.” (Ravazo-Rodriguez  et al., 2017:1) With an estimation of three errors out of 21 evaluations this would have indicated a roughly 85% accuracy rate and, as I said above this is not bad in a field like art history, perhaps not good enough for a scientist, but encouraging for me.

NOTE: Some images in this posting were retrieved from the internet with a search for public domain photographs. If any of these images are not intended to be public domain, I apologize, and will happily provide the picture credits if the owner will contact me with them. For further information on this you should read the original reports at the sites listed below.

REFERENCES:

Coffey, Rebecca, 2012, Digit Divide, Scientific American, July 2012, p. 19.

Faris, Peter, 2009, Hand prints in Rock Art, 5 August 2009, RockArtBlog, https://rockartblog.blogspot.com.

Messer, A’ndrea Elyse, 2013, Women leave their handprints on the cave wall, 15 October 2013, Penn State University press release. https://www.psu.edu/news/research/story/women-leave-their-handprints-cave-wall. Accessed online 14 April 2025.

Ravazo-Rodriguez, Ana Maria et al., 2017, New data on the sexual dimorphism of the hand stencils in El Castillo Cave (Cantabria, Spain), Journal of Archaeological Science: Reports 14 (2017), 374-381.

SECONDARY REFERENCE:

Snow, D.R., 2013. Sexual dimorphism in European Upper Paleolithic cave art. Am. Antiq. 78 (4), 746–761.

Wang, James Z. et al., 2013, Determining the Sexual Identities of Prehistoric Cave Artists using Digitized Handprints, A Machine Learning Approach, Penn State University. Downloaded from Research Gate on 7 September 2025.

 

HANDEDNESS:

On August 5, 2009, I posted a column on Hand prints in Rock Art in which I discussed the fact that a viewer can sometimes determine the gender of a rock art creator by measuring the relative length of the first and third fingers in a hand print. Statistically more males have a longer third finger while more females have a longer first finger.
  

I believe it is also possible to get a suggestion of the handedness of the creator of the rock art by examining a hand print. A human normally finds one side or the other dominant and capable of finer control and dexterity.

Hand-prints, Texas. Photo: Peter Faris, 2004.

According to Wikipedia right-handedness is most common. Right-handed people are more dexterous with their right hands when performing tasks. A variety of studies suggest that 70-90% of the world population is right-handed, rather than left-handed or any other form of handedness. Left-handedness is less common than right-handedness. Left-handed people are more dexterous with their left hands when performing tasks. A variety of studies suggest that 8–15% of the world population is left-handed.

Now how would we apply this knowledge to the analysis of a hand print in rock art? It seems to me that if the hand print is actually a print of a hand – a hand covered in paint and applied flat to the rock face – the creator of the image would probably use their dominant hand out of force of habit. Therefore if it is a print of a right hand it was probably a right-handed person and if it is a print of a left hand it was probably a left-handed person.

If, on the other hand (pun intended), it is a tracing of a hand, whether painted or pecked, it is probably the opposite because the creator would be using their dominant hand to do the tracing around the less useful hand that they were holding flat against the rock face. When the outline was made permanent, either with paint or by pecking, it was a recording of the person’s subordinate hand, not the dominant one.

Applying this reasoning to the same panel of hand prints in Texas that I photographed in 2004 and used to illustrate the August 5, 2009 posting about hand-prints we see that these hand-prints were made with a paint covered right hand applied flat to the rock face. We have already seen that these prints indicated that their creator was probably a male, now we can see that he was probably a right-handed male. Not bad for a group of anonymous hand-prints.

HANDPRINTS IN ROCK ART:

Hand prints, Pecos trip, Texas, Jan. 2004.

I believe that all students of rock art are fascinated by the symbol of the hand print – I certainly find them intriguing. They come from all ages, from Paleolithic caves in Europe, Aboriginal rock shelters in Australia, and North American rock art panels from all periods. They have been created in many forms; pictographic and petroglyphic, and in all degrees of realism. Some of the pictographic versions seem to have been made by pressing a painted hand against the rock. Some others found have been spray painted by blowing a mouthful of paint around the hand carefully held flat against the rock. Other hand prints are outline or solidly pecked petroglyphs. It is common to think of them, especially the painted hand stamps, as a sort of a signature by the creator of the image, a statement that “I was here.” This is an example of interpretation through empathy. We believe this because it makes sense in our common humanity with the artists. We see it as something they would do because it is something that we would do. We can also think of them in this sense because many of them reveal individual, identifiable traits. In some cases variations in size may determine that some hand prints were left by children or adolescents. Many Paleolithic examples in some of the caves of Europe show missing fingers or joints. These are speculated to be victims of frostbite, or cases of digits cut off in grief as in known examples from the Great Plains tribes of North America so many years later.

Larry Loendorf, in his recent book Thunder and Herds wrote that “adult human beings seem not to have abandoned the innate fascination with their hands. In the rock art record from the earliest to the most recent of times the hand is the most common stylistic element. Red hand stencils and prints occur in panels dated between 30,000 and 36,000 years ago at Chauvet cave in France. They occur in large numbers in complex, painted arrangements in caves in Borneo and Argentina, and as single petroglyphs – associated with spirals – at sites at Zuni and La Cienega, New Mexico.”

The more realistic, exact and accurate, hand prints, whether painted by hand stamping, spraying, or pecking after outlining the person’s real hand on the stone, bring up another fascinating possibility – determining the gender of the creator of the image. There is an interesting difference between the hands of most men and most women. Statistically, most male hands have a 3rd (or ring) finger that is longer than the 1st (index) finger. At the same time, statistically, more women show the opposite, an index finger that is longer than the ring finger. Not all, this is not a universal trait, but the larger proportion of a study group should show this.

In the example shown from the Pecos region in Texas, the 3rd finger appears noticeably longer than the 1st finger to me. Therefore, I interpret this to mean that the maker of the hand prints was probably a male. A researcher at the site could perform careful measurements to prove this one way or another. Another possible way of adding to our knowledge about the makers of the rock art.